Fig. 1. Sequence logos of RING and B-box motifs and their domain co-occurrence with RBL subtypes. (a) Multiple sequence alignments illustrated with sequence logos. Red indicates hydrophobic residues IVLFCMA (4.5 to 1.8 on the Kyte–Doolittle scale³⁴), green hydrophilic residues RKNDQEH (− 4.5 to − 3.2), and black intermediate residues GTWSYP (− 0.4 to − 1.6). (b) Venn diagrams showing the domain co-occurrences for the data sources UniprotKB/Swissprot UniprotKB, and a set of 420 genomes. For RING and B-box, Pfam HMMs were used in the search, using the trusted cutoff scores (TC parameter in Pfam HMMER). RBL-HMM was based on the in-house-built HMM for the respective subtype.

Fig. 2. Sequence logos and helical wheels for RBL subtypes. (a) Sequence logos show the conserved asparagine N-terminal to the α-helix in all three RBL subtypes together with subtype-specific conservation patterns, as described in Results, together with secondary structure prediction and corresponding reliability scores. (b) Helical wheels for predicted helical residues show amphipathic properties for all subtypes. Hydrophobicity values are the average of the Kyte–Doolittle scores for the respective position in the multiple sequence alignments. Sequence residues are from human TRIM2 (type A), human TRIM21/Ro52 (type B), and mouse TRIM50 (type C).

Fig. 3. (a) Ro52 constructs used in this study. Gray boxes correspond to the known conserved domains, with residue numbers as in Ro52 and where the RING–RBL and RBL–B-box constructs contain residues 1–91 and 55–128, respectively. (b) The fluorescence emission recorded at the maximum emission wavelength as a function of Zn²⁺ concentration (open circles) and the corresponding CaLigator³⁶ fit (solid line) that gives the binding constants indicated in the text. The values are the averages of three measurements, and deviations are represented by error bars. Each value was taken at equilibrium. (c) Experimental and calculated CD spectra of Ro52–RBL together with the secondary structure content estimation of the RBL in different contexts (inset; α, α-helix; rc, random coil). Spectra include experimental CD spectrum of the single RBL peptide (blue short dash), calculated spectrum of the RBL together with the B-box (red, line) and together with the RING (green long dash), as well as experimental spectra of the RING–RBL in the absence (pink small dots) and in the presence (black large dots) of Zn²⁺, respectively. (d) Antigenicity of the RING–RBL construct compared to the RBL–B-box under different conditions. Sera 1–4 are from healthy (negative) controls, and sera 5–13 are from patients with Sjögren syndrome. Anti-His denotes positive control with anti-His-tag antibodies. The different conditions are shown after each construct identifier: red, reduced; ox, oxidized; + zn, the construct in the Zn²⁺-bound state; − zn, the construct in the Zn²⁺-free state. The color code of OD (normalized values) is given to the right of the figure.

Fig. 4. (a) Analytical ultracentrifugation results for the RING–RBL. The circles in the lower main panel show the absorbance versus radius data for seven different centrifugation speeds (see inset) at equilibrium. The calculated fit is represented by continuous lines. The upper panel shows representative residuals at 42,000 rpm. (b) Stability measurements of the Ro52 RING–RBL. Fluorescence spectroscopy data are shown as fractional changes of emission maxima wavelength as a function of increasing GdmCl concentration in the absence (triangle) and in the presence of 1 Eq (circle) and 2 Eq (square) of Zn²⁺, respectively. Solid lines correspond to fits calculated by TableCurve 2D.

Fig. 5. (a) Proteolytic digestion time courses for Ro52 RING–RBL fragments derived from MALDI-TOF-MS data. The first row of profiles shows the degradation of full-length RING–RBL by three different proteases in the presence and in the absence of Zn²⁺. The second row shows the accumulation of stable RING–RBL fragments and degradation of larger or unstable fragments during proteolytic cleavage. The first column of the second row shows N-terminal fragments, the second column shows internal fragments, and the last column shows C-terminal fragments. The used protease is written above each panel. Solid lines correspond to samples without Zn²⁺, and dashed lines correspond to samples with Zn²⁺. (b) Cleavage propensities in the Ro52 RING–RBL protein as detected by MALDI-TOF-MS in the limited proteolysis experiment with and without Zn²⁺ as relative cleavage propensity 3D plots. Cleavage sites are labeled with their residue number as in the native sequence and framed with colored boxes according to the protease used (red, trypsin; blue, chymotrypsin; green, V8). A corresponding schematic domain overview of Ro52 is given below the propensity plots.

Fig. 6. The three different RING–Ro52–RBL–Ro52 structural models retrieved from docking experiments. (a) The final model based on AIRs derived from proteolysis data together with sequence homology, followed by models based on AIRs derived from sequence homology alone using hydrophobic (b) and hydrophilic (c) patches (blue, RING; red, RBL). Amino acid side chains that build the interaction interface are shown in green. In all three representations, the RING has the same orientation. Figures were generated with PyMOL [http://www.pymol.org].⁴¹

Evaluation of domainwise sequence identity

For each domain, the mean, median, and standard deviation of the percentage sequence identity were calculated for type A, type B, type C, and unclassified, respectively, each set versus itself and versus the others. The sequence identity is retrieved by running an all-against-all alignment comparison using water from the EMBOSS³⁵ package. The difference in sequence identity between the sets was evaluated using a heteroscedastic Student's t test. There are significant differences between all subtypes for all three domains.

A summary of the structural statistics of docking experiments

All values are averages of all dockings performed, which correspond to the description given in the first column.

Active and passive AIRs chosen for the docking experiments

Counting starts from the N-terminal residue of each model. In the last row, the active and passive AIRs, which were picked based on conservation and limited proteolysis data, are listed; thus, the terms hydrophobic and hydrophilic are not valid, since a mixture of both comprises these sets of AIRs.